Delay Vector Variance Method

The Delay Vector Variance (DVV) method characterises a time series in a
standardised way on the basis of its predictability in phase space, by
means of a so-called `DVV-plot'. By itself, these DVV-plots can be
used for time series clustering, as has been shown in Gautama et al.,
2003a for EEG signals. In the context of signal nonlinearity testing,
it can be used in combination with the `surrogate data' method (e.g.,
Schreiber and Schmitz, 2000), namely by characterising the `original'
time series and a number of `surrogates', and statistically testing
whether they are different (Gautama et al., 2003b).